Overview of BH3 profile
Junyan Lu
Load and preprocess BH3 profiling data
Load
Use baseline level from DBP profiling
Prepare sample background annotations
Data structure
Data dimension
Number of samples
## [1] 73Overview of CytC release data
For each concentration
For AUC
Dose response curves
Hierarchical clustering and heatmaps
Pricipal component analysis
PCA biplot
Associations between PCs and patient background
Associations with P-value < 0.05
## # A tibble: 4 × 3
## # Groups: PC, feature [4]
## PC feature p.value
## <chr> <chr> <dbl>
## 1 PC2 NOTCH1 0.00129
## 2 PC2 IGHV.status 0.00286
## 3 PC1 trisomy12 0.0102
## 4 PC2 Methylation_Cluster 0.0356Plot associations
Plot feature loadings on the first three PCs
Feature correlations
AUC
Individual concentrations
Association with the spontaneous apoptosis of CLL cells
Measured by image data
PCs
Individual peptide
If multiple concentrations are identified as significant, only show
the most significant concentration.
Association test with patient genomic background
Prepare patient genomic background
## [1] "IGHV.status" "del11q" "del13q" "del17p" "trisomy12" "NOTCH1" "SF3B1" "TP53"Plot to summarise genomic background
Association test
Test for Genomics
Methylation cluster
Table of associations
P value scatter plot
Box plots for the significant associations (10% FDR)
If multiple concentrations are identified as significant, only show
the most significant concentration.
Genomic correlation with individual concentrations
Association test
Summary heatmap plot for all concentrations
Associations with pretreatment status
Test for Genomics
## # A tibble: 7 × 4
## feature p.value estimate p.adj
## <chr> <dbl> <dbl> <dbl>
## 1 ABT199 0.0174 11.7 0.122
## 2 BAD 0.0460 9.51 0.161
## 3 BIM 0.414 2.99 0.580
## 4 FS1 0.190 4.23 0.333
## 5 HRKy 0.946 -0.0745 0.946
## 6 MS1 0.787 1.49 0.918
## 7 PUMA 0.116 6.85 0.270
##
## Pearson's Chi-squared test with Yates' continuity correction
##
## data: patAnno$IGHV.status and patAnno$pretreat
## X-squared = 4.9003, df = 1, p-value = 0.02685Estimate the confounding effect of pretreatment status
How many treated and untreated samples?
##
## no yes
## 55 18Adjust for pretreatment status
Test for Genomics
Methylation cluster
Combine
Test only in untreated
Test for Genomics
Methylation cluster
Table for comparing results
Association with transcriptomics
Preprocessing
RNAseq
BH3 profiling
Association test for each feature
Number of significant associations per feature (10% FDR)
## # A tibble: 7 × 2
## feature n
## <chr> <int>
## 1 ABT199 110
## 2 BAD 64
## 3 BIM 0
## 4 FS1 0
## 5 HRKy 0
## 6 MS1 0
## 7 PUMA 0Table of significant associations
Heamap for significant correlations
Plot top 9 significant associations
Plot top 9 significant associations for ABT199
Plot selected associations for ABT199
Plot selected associations for BAD
Combine plot for ABT199 and BAD
Pathway enrichment analysis for features that associate with gene expression
Cancer Hallmark
KEGG
Record siginificant RNAs for later feature selection
Gene set heatmap
Association with proteomics
Preprocessing
Proteomics
## [1] 3314 30BH3 profiling
Association test for each feature
Number of significant associations per feature (10% FDR)
None passed 10% FDR
Table of significant associations (P<0.01)
Plot top 9 associations
Multivariate feature selection for explaining BH3 profling in CLL
Data pre-processing
BH3 profiling
RNAseq
For genomic data
For demographic and clinical data
Function to Generate the explanatory dataset for each BH3 profile
Clean and integrate multi-omics data
Perform multi-variate regression
LASSO model
Training models
Function for multi-variate regression
Perform lasso regression
Ploting heatmap of selected features
Function for the heatmap plot
Plot all heatmaps
